DISTINCTCOUNTHLLPLUS

This section contains reference documentation for the DISTINCTCOUNTHLLPLUS function.

Returns an approximate distinct count using HyperLogLogPlusPlus. It also takes an optional second and third arguments to configure the p, sp for the HyperLogLogPlus. The optional parameter p defines the normal set precision and the parameter sp defines the sparse set precision. For accurate distinct counting, see DISTINCTCOUNT.

Signature

DISTINCTCOUNTHLLPlus(colName) DISTINCTCOUNTHLLPlus(colName, p) DISTINCTCOUNTHLLPlus(colName, p, sp)

Usage Examples

These examples are based on the Batch Quick Start.

select DISTINCTCOUNTHLLPLUS(teamID) AS value
from baseballStats 

value
158

select DISTINCTCOUNTHLLPLUS(teamID, 12) AS value
from baseballStats 

value
149

DISTINCTCOUNTSMARTHLLPLUS

DISTINCTCOUNTSMARTHLLPLUS starts with exact distinct counting in a set and switches to HyperLogLog++ when the number of distinct values exceeds a threshold. This is useful when some groups stay small enough for exact counting while larger groups still need bounded memory usage.

Signature

DISTINCTCOUNTSMARTHLLPLUS(colName) DISTINCTCOUNTSMARTHLLPLUS(colName, 'threshold=;p=

;sp=')

Parameters

• colName (required): Column to aggregate.

• threshold (optional): Number of distinct values Pinot keeps exactly before converting to HLL++. Default: 100000. A non-positive value disables the conversion.

• p (optional): HyperLogLog++ normal precision after conversion. Default: 14.

• sp (optional): HyperLogLog++ sparse precision after conversion. Default: 0.

Usage Examples

SELECT DISTINCTCOUNTSMARTHLLPLUS(teamID) AS value
FROM baseballStats

SELECT DISTINCTCOUNTSMARTHLLPLUS(teamID, 'threshold=10000;p=12') AS value
FROM baseballStats